Skip to contentSkip to conversation

drivers: add lcd display support

PATCH In review Miguel Ojeda Sandonis 2 participants 5 messages Sep 22, 2006 – Sep 28, 2006

Subsystems (6)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drivers: add lcd display support

(This is the real V5 patch. The other V5-mail was the V4, sorry)

miguelojeda-2.6.18-add-lcd-display-support.patch

Changelog (newer first):

V5 - Fixed mistakes found by Joe Perches
- coding style
- const, spelling mistake

V4 - Fixed mistakes found by Pavel Machek
- no zero initialization
- GPL version
- use bitop generic functions

V3 - Apply for 2.6.18
- added const to class_create (2.6.18 hasn't add-const-to-class_create.patch)

V2 - Fixed mistakes found by Greg KH
- coding style

V1 - Fixed mistakes found by Alexey Dobriyan
- coding style
- init() and exit() error management

V0 - Original patch

Brief:

- Adds const to class_create (2.6.18 hasn't add-const-to-class_create.patch)
- Adds a "lcddisplay" class for registering LCD devices.
- Adds support for the ks0108 LCD Controller as a device driver.
(uses lcddisplay class and parport interface)
- Adds support for the cfag12864b LCD Display as a device driver.
(uses ks0108 device driver)
- Adds a new ioctl() magic number/range (0xFF) in the list for the cfag12864b device.
- Adds the usual Documentation, includes, Makefiles, Kconfigs, MAINTAINERS, CREDITS...

Patched files Index:

patching file drivers/base/class.c
patching file include/linux/device.h
patching file drivers/Kconfig
patching file drivers/lcddisplay/cfag12864b.c patching file drivers/lcddisplay/cfag12864b_image.h patching file drivers/lcddisplay/Kconfig patching file drivers/lcddisplay/ks0108.c patching file drivers/lcddisplay/lcddisplay.c patching file drivers/lcddisplay/Makefile patching file drivers/Makefile
patching file include/linux/cfag12864b.h patching file include/linux/ks0108.h
patching file include/linux/lcddisplay.h patching file Documentation/drivers/lcddisplay/cfag12864b patching file Documentation/drivers/lcddisplay/lcddisplay patching file Documentation/ioctl-number.txt patching file CREDITS
patching file MAINTAINERS

Please review.

miguelojeda-2.6.18-add-lcd-display-support.patch

---

drivers/base/class.c +1 -1
diff -uprN -X dontdiff linux-2.6.18-vanilla/drivers/base/class.c linux-2.6.18/drivers/base/class.c
@@ -197,7 +197,7 @@ static int class_device_create_uevent(st
  * Note, the pointer created here is to be destroyed when finished by
  * making a call to class_destroy().
  */
-struct class *class_create(struct module *owner, char *name)
+struct class *class_create(struct module *owner, const char *name)
 {
 	struct class *cls;
 	int retval;
include/linux/device.h +1 -1
diff -uprN -X dontdiff linux-2.6.18-vanilla/include/linux/device.h linux-2.6.18/include/linux/device.h
@@ -271,7 +271,7 @@ struct class_interface {
 extern int class_interface_register(struct class_interface *);
 extern void class_interface_unregister(struct class_interface *);
 
-extern struct class *class_create(struct module *owner, char *name);
+extern struct class *class_create(struct module *owner, const char *name);
 extern void class_destroy(struct class *cls);
 extern struct class_device *class_device_create(struct class *cls,
 						struct class_device *parent,
drivers/Kconfig +2 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/drivers/Kconfig linux-2.6.18/drivers/Kconfig
@@ -74,4 +74,6 @@ source "drivers/rtc/Kconfig"
 
 source "drivers/dma/Kconfig"
 
+source "drivers/lcddisplay/Kconfig"
+
 endmenu
drivers/lcddisplay/cfag12864b.c +529 -0

529 lines not on this page · raw message

drivers/lcddisplay/cfag12864b_image.h +80 -0

80 lines not on this page · raw message

drivers/lcddisplay/Kconfig +61 -0

61 lines not on this page · raw message

drivers/lcddisplay/ks0108.c +175 -0

175 lines not on this page · raw message

drivers/lcddisplay/lcddisplay.c +71 -0

71 lines not on this page · raw message

drivers/lcddisplay/Makefile +7 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/drivers/lcddisplay/Makefile linux-2.6.18/drivers/lcddisplay/Makefile
@@ -0,0 +1,7 @@
+#
+# Makefile for the kernel LCD Display device drivers.
+#
+
+obj-$(CONFIG_LCDDISPLAY)	+= lcddisplay.o
+obj-$(CONFIG_KS0108)		+= ks0108.o
+obj-$(CONFIG_CFAG12864B)	+= cfag12864b.o
drivers/Makefile +1 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/drivers/Makefile linux-2.6.18/drivers/Makefile
@@ -76,3 +76,4 @@ obj-$(CONFIG_CRYPTO)		+= crypto/
 obj-$(CONFIG_SUPERH)		+= sh/
 obj-$(CONFIG_GENERIC_TIME)	+= clocksource/
 obj-$(CONFIG_DMA_ENGINE)	+= dma/
+obj-$(CONFIG_LCDDISPLAY)	+= lcddisplay/
include/linux/cfag12864b.h +45 -0

45 lines not on this page · raw message

include/linux/ks0108.h +21 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/include/linux/ks0108.h linux-2.6.18/include/linux/ks0108.h
@@ -0,0 +1,21 @@
+/*
+ *    Filename: ks0108.h
+ *     Version: 0.1.0
+ * Description: ks0108 LCD Display Controller Driver Header
+ *     License: GPLv2
+ *
+ *      Author: Miguel Ojeda Sandonis <[email protected]>
+ *        Date: 2006-09-12
+ */
+
+#ifndef _KS0108_H_
+#define _KS0108_H_
+
+extern void ks0108_writedata(unsigned char byte);
+extern void ks0108_writecontrol(unsigned char byte);
+extern void ks0108_displaystate(unsigned char state);
+extern void ks0108_startline(unsigned char startline);
+extern void ks0108_address(unsigned char address);
+extern void ks0108_page(unsigned char page);
+
+#endif /* _KS0108_H_ */
include/linux/lcddisplay.h +19 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/include/linux/lcddisplay.h linux-2.6.18/include/linux/lcddisplay.h
@@ -0,0 +1,19 @@
+/*
+ *    Filename: lcddisplay.h
+ *     Version: 0.1.0
+ * Description: Display Class Header
+ *     License: GPLv2
+ *
+ *      Author: Miguel Ojeda Sandonis <[email protected]>
+ *        Date: 2006-09-12
+ */
+
+#ifndef _LCDDISPLAY_H_
+#define _LCDDISPLAY_H_
+
+#include <linux/device.h>
+
+extern struct class * lcddisplay_class;
+
+#endif /* _LCDDISPLAY_H_ */
+
Documentation/drivers/lcddisplay/cfag12864b +367 -0

367 lines not on this page · raw message

Documentation/drivers/lcddisplay/lcddisplay +37 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/Documentation/drivers/lcddisplay/lcddisplay linux-2.6.18/Documentation/drivers/lcddisplay/lcddisplay
@@ -0,0 +1,37 @@
+	=================================
+	LCD Display Drivers Documentation
+	=================================
+
+License:		GPL
+Author & Maintainer:	Miguel Ojeda Sandonis <[email protected]>
+Date:			2006-09-12
+
+--------
+0. INDEX
+--------
+
+	1. NEW DISPLAY DRIVERS
+	2. GENERAL TIPS
+
+
+
+----------------------
+1. NEW DISPLAY DRIVERS
+----------------------
+
+Feel free to send me new display drivers. I will try to do my best.
+
+If you don't get any answer, send your patch directly to the linux-kernel ml.
+
+
+
+---------------
+2. GENERAL TIPS
+---------------
+
+- Divide your driver into the controller driver, like ks0108,
+  and the specific LCD display series driver, like cfag12864b.
+
+- Claim for your IO ports in the controller driver.
+
+EOF
Documentation/ioctl-number.txt +2 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/Documentation/ioctl-number.txt linux-2.6.18/Documentation/ioctl-number.txt
@@ -191,3 +191,5 @@ Code	Seq#	Include File		Comments
 					<mailto:[email protected]>
 0xF3	00-3F	video/sisfb.h		sisfb (in development)
 					<mailto:[email protected]>
+0xFF	00-1F	linux/cfag12864b.h	cfag12864b LCD Display Driver
+					<mailto:[email protected]>
CREDITS +8 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/CREDITS linux-2.6.18/CREDITS
@@ -2535,6 +2535,14 @@ S: Subiaco, 6008
 S: Perth, Western Australia
 S: Australia
 
+N: Miguel Ojeda Sandonis
+E: [email protected]
+D: Author: LCD Display Drivers (ks0108, cfag12864b)
+D: Maintainer: LCD Display Drivers Tree (drivers/lcddisplay/*)
+S: C/ Mieses 20, 9-B
+S: Valladolid 47009
+S: Spain
+
 N: Greg Page
 E: [email protected]
 D: IPX development and support
MAINTAINERS +5 -0
diff -uprN -X dontdiff linux-2.6.18-vanilla/MAINTAINERS linux-2.6.18/MAINTAINERS
@@ -1707,6 +1707,11 @@ M:	[email protected]
 L:	[email protected]
 S:	Maintained
 
+LCD DISPLAY DRIVERS
+P:	Miguel Ojeda Sandonis
+M:	[email protected]
+S:	Maintained
+
 LED SUBSYSTEM
 P:	Richard Purdie
 M:	[email protected]
#1 lore raw

include/linux/device.h struct class_interface { · line 274

Quoted 13 lines · On Fri, 22 Sep 2006 22:03:46 +0200 Miguel Ojeda Sandonis…

miguelojeda-2.6.18-add-lcd-display-support.patch

...

diff -uprN -X dontdiff linux-2.6.18-vanilla/include/linux/device.h linux-2.6.18/include/linux/device.h

include/linux/device.h +1 -1
@@ -271,7 +271,7 @@ struct class_interface {
 extern int class_interface_register(struct class_interface *);
 extern void class_interface_unregister(struct class_interface *);
 
-extern struct class *class_create(struct module *owner, char *name);
+extern struct class *class_create(struct module *owner, const char *name);

Please prepare a separate patch for this, send to Greg.

Quoted 18 lines
drivers/lcddisplay/cfag12864b.c +15 -0
@@ -0,0 +1,529 @@
+/*
+ *    Filename: cfag12864b.c
+ *     Version: 0.1.0
+ * Description: cfag12864b LCD Display Driver
+ *     License: GPLv2
+ *     Depends: lcddisplay ks0108
+ *
+ *      Author: Miguel Ojeda Sandonis <[email protected]>
+ *        Date: 2006-09-21
+ */
+
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/kernel.h>

drivers/lcddisplay/cfag12864b.c line 25

Quoted 10 lines
drivers/lcddisplay/cfag12864b.c +10 -0
+#include <linux/fs.h>
+#include <linux/cdev.h>
+#include <linux/device.h>
+#include <linux/lcddisplay.h>
+#include <linux/ks0108.h>
+#include <linux/cfag12864b.h>
+#include <asm/uaccess.h>
+
+#define NAME "cfag12864b"
+#define PRINTK_PREFIX  KERN_INFO NAME ": "

In numerous places the driver uses PRINTK_PREFIX for reporting errors. It should use KERN_ERR. I suggest that PRINTK_PREFIX be removed and that you open-code all the printk facility levels, make sure that the appropriate level is being used at each site.

drivers/lcddisplay/cfag12864b.c line 35

Quoted 9 lines
drivers/lcddisplay/cfag12864b.c +9 -0
+
+
+/*
+ * Device
+ */
+
+static const unsigned int cfag12864b_firstminor;
+static const unsigned int cfag12864b_ndevices = 1;
+static const char * cfag12864b_name = NAME;

The more usual kernel style is to not put a space after the `*':

	static const char *cfag12864b_name = NAME;

(this affects the entire patch).

Suggest that cfag12864b_name be removed - just use NAME at the single place where this variable is used.

drivers/lcddisplay/cfag12864b.c line 43

Quoted 8 lines
drivers/lcddisplay/cfag12864b.c +8 -0
+static int cfag12864b_major;
+static int cfag12864b_minor;
+static dev_t cfag12864b_device;
+struct cdev cfag12864b_chardevice;
+
+
+
+

One blank line is sufficient (there are many instances of this)

Quoted 17 lines
drivers/lcddisplay/cfag12864b.c +17 -0
+/*
+ * cfag12864b Commands
+ */
+
+static const unsigned int cfag12864b_bits = 8;
+static const unsigned int cfag12864b_width = CFAG12864B_WIDTH;
+static const unsigned int cfag12864b_height = CFAG12864B_HEIGHT;
+static const unsigned int cfag12864b_matrixsize = CFAG12864B_MATRIXSIZE;
+static const unsigned int cfag12864b_controllers = CFAG12864B_CONTROLLERS;
+static const unsigned int cfag12864b_pages = CFAG12864B_PAGES;
+static const unsigned int cfag12864b_addresses = CFAG12864B_ADDRESSES;
+static const unsigned int cfag12864b_size = CFAG12864B_SIZE;
+static unsigned char cfag12864b_state;
+
+static void cfag12864b_set(void)
+{
+	ks0108_writecontrol(cfag12864b_state);

drivers/lcddisplay/cfag12864b.c line 70

Quoted 10 lines
drivers/lcddisplay/cfag12864b.c +10 -0
+}
+
+static void cfag12864b_setbit(unsigned char state, unsigned char bit)
+{
+	if (state)
+		set_bit(bit, (void*)&cfag12864b_state);
+	else
+		clear_bit(bit, (void*)&cfag12864b_state);
+	cfag12864b_set();
+}

bitops are defined on an unsigned long only. This trick is as ugly as sin and is buggy on big-endian CPUs. Suggest that cfag12864b_state be converted to unsigned long.

drivers/lcddisplay/cfag12864b.c line 159

Quoted 6 lines
drivers/lcddisplay/cfag12864b.c +6 -0
+{
+	cfag12864b_startline(0);
+}
+
+
+

blank lines

drivers/lcddisplay/cfag12864b.c line 164

Quoted 5 lines
drivers/lcddisplay/cfag12864b.c +5 -0
+/*
+ * Auxiliary
+ */
+
+static void normalizeoffset(unsigned int * offset)

coding style.

drivers/lcddisplay/cfag12864b.c line 168

Quoted 4 lines
drivers/lcddisplay/cfag12864b.c +4 -0
+{
+	if (*offset >= cfag12864b_pages*cfag12864b_addresses)
+		*offset -= cfag12864b_pages*cfag12864b_addresses;
+}

Ths usual kernel style is to put a single space around arithmetic operators and around comparison operators. This affects the entire patch, please.

drivers/lcddisplay/cfag12864b.c line 212

Quoted 6 lines
drivers/lcddisplay/cfag12864b.c +6 -0
+void cfag12864b_clear(void)
+{
+	unsigned char page,address;
+
+	cfag12864b_setcontrollers(1, 1);
+	for (page=0; page<cfag12864b_pages; ++page) {

For example,

	for (page = 0; page < cfag12864b_pages; page++) {

would be more kernelish.

The use of the identifier page' here is unfortunate. We usually expect such a thing to have type struct page *'. I understand that "Each controller is divided into 8 pages", but it'd be nice if some different nomenclature could be used here. If nothing else comes to mind, we can live with it as-is.

drivers/lcddisplay/cfag12864b.c line 224

Quoted 5 lines
drivers/lcddisplay/cfag12864b.c +5 -0
+void cfag12864b_write(
+	unsigned short offset,
+	const unsigned char * buffer,
+	unsigned short count)
+{

It is more usual to do

void cfag12864b_write(unsigned short offset, const unsigned char *buffer,
unsigned short count)
{

drivers/lcddisplay/cfag12864b.c line 286

Quoted 8 lines
drivers/lcddisplay/cfag12864b.c +8 -0
+	for (controller=0; controller < cfag12864b_controllers; ++controller)
+	for (page=0; page < cfag12864b_pages; ++page)
+	for (address=0; address < cfag12864b_addresses; ++address) {
+		dest[(controller*cfag12864b_pages+page)*cfag12864b_addresses+address]=0;
+		for (bit=0; bit < cfag12864b_bits; ++bit)
+			if (src[controller*cfag12864b_addresses+address+(page*cfag12864b_bits+bit)*cfag12864b_width])
+				set_bit(bit, (void*)(dest+(controller*cfag12864b_pages+page)*cfag12864b_addresses+address));
+	}

That's rather ugly-looking.

It's also probably-incorrect on big-endian CPUs. Perhaps you should not use bitops at all for this driver, use open-coded | and &/~ instead?

The driver doesn't have any locking. Is it racy on SMP and/or CONFIG_PREEMPT?

Quoted 10 lines
drivers/lcddisplay/cfag12864b.c +10 -0
+static int cfag12864b_fopioctlformat(void __user * arg)
+{
+	int result;
+	int ret = -ENOTTY;
+
+	unsigned char * tmpbuffer;
+
+	tmpbuffer = kmalloc(
+		sizeof(unsigned char)*cfag12864b_matrixsize,GFP_KERNEL);
+	if (tmpbuffer == NULL) {

drivers/lcddisplay/cfag12864b.c line 394

Quoted 10 lines
drivers/lcddisplay/cfag12864b.c +10 -0
+		printk(PRINTK_PREFIX "FOP ioctl: ERROR: "
+			"can't alloc memory %i bytes\n",
+			sizeof(unsigned char)*cfag12864b_matrixsize);
+		goto none;
+	}
+
+	result = copy_from_user(
+		tmpbuffer,
+		arg,
+		sizeof(unsigned char)*cfag12864b_matrixsize);
	result = copy_from_user(tmpbuffer, arg, cfag12864b_matrixsize);
Quoted 12 lines
drivers/lcddisplay/cfag12864b.c +12 -0
+	if (result != 0) {
+		printk(PRINTK_PREFIX "FOP ioctl: ERROR: "
+			"can't copy memory from user\n");
+		goto bufferalloced;
+	}
+	
+	cfag12864b_format(tmpbuffer);
+
+	ret = 0;
+
+bufferalloced:
+	kfree(tmpbuffer);

drivers/lcddisplay/cfag12864b.c line 443

Quoted 10 lines
drivers/lcddisplay/cfag12864b.c +7 -0
+
+none:
+	return ret;
+}
+
 
...
 
+
+static struct file_operations cfag12864b_fops =

`static const struct'

drivers/lcddisplay/cfag12864b.c line 498

Quoted 5 lines
drivers/lcddisplay/cfag12864b.c +5 -0
+
+	class_device_create(
+		lcddisplay_class,NULL,
+		cfag12864b_device,
+		NULL,"cfag12864b%d", cfag12864b_minor);

class_device_create() can fail.

drivers/lcddisplay/cfag12864b.c

Quoted 6 lines
drivers/lcddisplay/cfag12864b.c +6 -0
+void ks0108_writecontrol(unsigned char byte)
+{
+	const unsigned int ecycledelay = 2;
+	udelay(ecycledelay);
+	parport_write_control(ks0108_parport, byte^(bit(3)|bit(1)|bit(0)));
+}

udelay(2) would be clearer.

drivers/lcddisplay/cfag12864b.c

Quoted 5 lines
drivers/lcddisplay/cfag12864b.c +5 -0
+void ks0108_displaystate(unsigned char state)
+{
+	unsigned char cmd = bit(1) | bit(2) | bit(3) | bit(4) | bit(5);
+	if (state)
+		set_bit(0, (void*)&cmd);

argh. Even if this driver will only ever run on big-endian hardware, please don't do this.

drivers/lcddisplay/cfag12864b.c

Quoted 6 lines
drivers/lcddisplay/cfag12864b.c +6 -0
+void ks0108_startline(unsigned char startline)
+{
+	const unsigned char maxstartline = 63;
+	unsigned char cmd = bit(6) | bit(7);
+	if (startline > maxstartline)
+		startline = maxstartline;

There are a lot of open-coded min() and max() operations in this driver. Suggest it be changed to use min() and max().

#2 lore raw
Quoted 4 lines · Andrew Morton

It's also probably-incorrect on big-endian CPUs. Perhaps you should not
use bitops at all for this driver, use open-coded | and &/~ instead?

Uhm, someone told me that they shouldn't be used because the kernel has generic functions for that.

I researched the kernel sources, and looking at bitops.h I found setbit(), so I tried to use it in the driver, althought I prefer standard |= and &=.

Are there more generic bitops kernel functions I may didn't find and I should use in the future?

Anyway, it's just a question: I will turn back the driver to use |= and &=, they look better to me too.

#3 lore raw
Quoted 8 lines · On Wed, 27 Sep 2006 20:58:24 +0000 "Miguel Ojeda" <[email protected]> wrote:
Quoted 4 lines · Andrew Morton

It's also probably-incorrect on big-endian CPUs. Perhaps you should not
use bitops at all for this driver, use open-coded | and &/~ instead?

Uhm, someone told me that they shouldn't be used because the kernel has generic functions for that.

You can't believe everything you read on the internet ;)

set_bit() and friends are a) atomic wrt other CPUs on SMP and b) only to be performed on unsigned longs.

Quoted 3 lines

I researched the kernel sources, and looking at bitops.h I found setbit(), so I tried to use it in the driver, althought I prefer standard |= and &=.

Those are the appropriate operations to use in this driver.

#4 lore raw
Quoted 14 lines · Andrew Morton
Quoted 8 lines · On Wed, 27 Sep 2006 20:58:24 +0000 "Miguel Ojeda" <[email protected]> wrote:
Quoted 4 lines · Andrew Morton

It's also probably-incorrect on big-endian CPUs. Perhaps you should not
use bitops at all for this driver, use open-coded | and &/~ instead?

Uhm, someone told me that they shouldn't be used because the kernel has generic functions for that.

You can't believe everything you read on the internet ;)

Quoted 10 lines

set_bit() and friends are a) atomic wrt other CPUs on SMP and b) only to be performed on unsigned longs.

Quoted 3 lines

I researched the kernel sources, and looking at bitops.h I found setbit(), so I tried to use it in the driver, althought I prefer standard |= and &=.

Those are the appropriate operations to use in this driver.

Ok, I have fixed all the mistakes you found. However, I have a question about locking.

I have used 1 global mutex for all the fops: seek, write and ioctl. They call down_interruptable() at the very beggining and up() just before returning. I think that it's right.

However, what about the exported symbols? I have exported functions at the ks0108 driver like EXPORT_SYMBOL_GPL(ks0108_writedata); The cfag12864b drivers exports more symbols also, like:
EXPORT_SYMBOL_GPL(cfag12864b_write);

Should use the mutex all this functions too?

For example: Another driver can have a fops which isn't perform any kind of locking. Such function is being executed twice at the same time. And it calls some of my exported symbols. Then, I will have a race condition in my driver.

(I couldn't find anything specific about that at LDD3, just for fops).

Thanks,
Miguel Ojeda